Children’s accuracy in answering Why and How Come questions

Abstract

Children’s developing understanding of language may influence their ability to accurately respond to questions inquiring about their event knowledge (i.e., Why and How Come questions), potentially creating misinterpretations in adult–child communication. The present study examined 120 5-, 7-, and 9-year-old’s accuracy in responding to Why and How Come questions about the cause of their behaviors. Children’s accuracy improved with age, highlighting a developmental milestone whereby children become highly accurate by 7 years of age. Further, the semantic differences in question type did not influence children’s responses, as there were no differences in children’s accuracy when answering Why or How Come questions. The findings from this study highlight the developmental shift in children’s abilities to answer Why and How Come questions, and thus the importance of considering the age and linguistic abilities of the child when inquiring about their event knowledge.

Children are regularly questioned by adults including parents, educators, medical professionals, social workers, and legal professionals (e.g., police officers, lawyers, judges), with the goal of obtaining information from children (Fritzley & Lee, 2003; Malloy & Stolzenberg, 2019). Furthermore, as noted by Fritzley and Lee (2003), asking children questions is the most common method used in developmental research, with 74% of studies (between 1995 and 1998) examining young children using questioning as part of their methodology. Therefore, it is important that children are able to accurately respond to questions asked of them, as children’s responses inform adults about the child, including their experiences, and can be used to assess children’s knowledge or understanding of those experiences. However, the accuracy of children’s responses depends not only on the child’s ability to produce the appropriate response but the adult’s ability to clearly communicate the question being asked (Saywitz et al., 1993). Therefore, it is critical to investigate questions that are potentially challenging for children or lack clarity in their request.

Much of our understanding on age appropriate questions for children comes from the child witness literature. Past research has largely focused on the benefits and challenges of wh- questions when questioning young children (e.g., Ervin-Tripp, 1970). When questioning child witnesses, best practices recommend the use of open-ended prompts (e.g., “Tell me more about [that].”) and wh-questions (e.g., who, what, when, where, and why questions) to obtain the most detailed and accurate reports (Lamb et al., 2018). Wh- questions can be of particular benefit, as they are narrow in requesting specific information, but unlike closed-ended questions (e.g., yes/no or forced choice questions), they allow the child to self-generate information. However, while wh- questions have generally been found to support young children in providing accurate and productive responses, some forms of wh- questions have been found to be more challenging than others (e.g., Andrews et al., 2016). Past research suggests that children’s acquisition of wh- questions develops in a chronological order, first using Who, What, and Where questions, followed by Why, How, and When questions in their own vocabulary (Ervin-Tripp, 1970; Tyack & Ingram, 1977). Furthermore, when responding to these wh- questions, researchers suggest that young children often experience greater difficulties with When, How, and Why questions as they ask about abstract concepts such as manner, causation, purpose, or time, compared to Who, What, and Where questions that ask about more concrete concepts (deVilliers & deVilliers, 1978).

One of the later acquired wh- questions commonly used in adult–child conversations includes Why (also phrased as How Come) questions, which can be particularly useful in determining whether children were coerced to perform an action (e.g., legal context) or the intention of a behavior (e.g., schoolyard misbehavior). Although semantically differentiated, the use of Why and How Come questions are interchangeable, as the intention of the questions is virtually the same. Within the field of forensic interviewing, research suggests that Why questions are frequently used when questioning young children (Ahern et al., 2018; Andrews et al., 2016; Malloy et al., 2016). For example, Malloy and colleagues found that of 49 investigative interviews, Why questions represented 9.3% of all interviewer utterances. However, although commonly asked, children are not always successful in responding to Why questions, as this ability develops with age. For example, Malloy and colleagues found that when examining forensic interviews with 3- to 5-year-olds, children’s responses were not always informative; providing the information sought by the interviewer’s Why questions only 20% of the time. However, Andrews and colleagues found that children 6 to 12 years old were able to provide productive and detailed responses to Why questions, demonstrating a developmental improvement in the productivity of children’s responses.

A possible explanation for younger children’s difficulty with Why/How Come questions is a lack of experience with these questions. Research suggests that children experience a lower frequency of Why questions within their everyday environment and therefore exhibit a later acquisition of this question type, compared to some other wh- questions (e.g., Who, What, Where; Ervin-Tripp, 1970; Rowland et al., 2003; Tyack & Ingram, 1977). In contrast, researchers also suggest that children experience fewer Why questions because they acquire these questions later in development (Cameron-Faulkner et al., 2003; Valian & Casey, 2003). That is, parents often exhibit a tendency to use questions that their children appear to understand; therefore, refraining from asking Why questions until children use these questions themselves, which occurs later in development relative to other wh- questions. Altogether, however, it remains clear that children have limited experience with Why questions, which may lead to difficulties, as children learn the social and pragmatic rules for answering questions through experience. It is also possible that younger children’s difficulties with these questions may arise due to their limited metalinguistic awareness of intentionality in communication exchanges, as this ability develops with age (Edwards & Kirkpatrick, 1999). Answering questions such as Why/How Come questions appropriately requires children to understand the linguistic intention of the question. Therefore, children who experience difficulties with Why/How Come questions may have yet to acquire the lexical knowledge necessary to understand the intentions of Why/How Come questions.

It is important to note that, as past research has focused on children’s responses to Why questions within field interviews, the accuracy of children’s responses could not be assessed as the accuracy of information is not known in these legal cases. This is particularly problematic, given that it is the accuracy, rather than quantity, of details children provide that reflect an understanding for the question and ability to respond appropriately. Therefore, the current study sought to examine the accuracy of children’s responses within a laboratory-based setting.

The Current Study

The current study examined 5-, 7-, and 9-year olds’ accuracy in answering Why and How Come questions about their own actions following a series of action-based events, with this age range capturing a period of increasing linguistic development and understanding of intentionality in communication exchanges (e.g., Edwards & Kirkpatrick, 1999). Furthermore, in response to Why questions, Malloy and colleagues found that 3- to 5-year-olds provided uninformative responses, while Andrews and colleagues found that 6- to 12-year-olds provided productive and detailed responses; therefore, the current age range was chosen to capture possible developmental changes in accuracy across these age groups. Given that children’s experience with Why/How Come questions and linguistic development increases with age (Cameron-Faulkner et al., 2003; Edwards & Kirkpatrick, 1999; Rowland et al., 2003; Valian & Casey, 2003), it was expected that older children would be more accurate, across question type (Why, How Come), compared to younger children. Additionally, past research has grouped Why and How Come questions together in analyses, as there were no differences in the productivity of their responses (e.g., Andrews et al., 2016). However, research suggests that minor differences in semantics can influence interpretations of the question and in turn the accuracy of children’s responses (e.g., Evans et al., 2014; Wylie et al., 2019). For example, How Come questions may express greater surprise or disbelief, whereas Why questions are more accusatory or critical which may lead children to justify their actions rather than accurately describe the cause of the event (Walker, 1999). Therefore, it was expected that children would be less accurate when answering Why questions compared to How Come questions. Furthermore, when children failed to provide an accurate response, we assessed whether they identified their lack of understanding (e.g., stated “I don’t know” [IDK]) or provided an inaccurate response. Given that past research suggests that children attempt to answer questions that they do not understand (Saywitz et al., 1993), we expected children to provide significantly more inaccurate responses compared to IDK responses overall.

Finally, given the important role of linguistic development in children’s ability to understand and answer Why and How questions (Edwards & Kirkpatrick, 1999), it was expected that children’s language ability would mediate the relationship between age and accuracy.

Method

Participants

One-hundred and twenty Canadian English-speaking children participated in this study (M_age = 89.38 months, SD = 20.07; 49% males): 40 5-year-olds (M_age = 65.3 months, SD = 3.64), 40 7-years-olds (M_age = 89.45 months, SD = 3.32), and 40 9-year-olds (M_age = 113.45 months, SD = 3.66). Participants were recruited from the community through a database of families who were interested in participating in research, as well as from a local Science Centre. Total household income was collected as an indicator of economic status, where the majority of households (58%) reported an income over CAD$75,000 (14% did not report total household income). Sixty-seven percent of participants identified as Caucasian, 7% as Asian, 4% as South-Asian, 2% as African-Canadian, 2% as Hispanic, 7% as other, and 11% did not report child ethnicity. This study was approved by the Brock University ethics board. Prior to beginning testing, written consent was obtained from all parents and verbal assent from all children.

Procedure

All children were tested individually in an on-campus research laboratory or in an area of a local Science Centre dedicated to testing participants. Children were led through five activity stations with the experimenter. Each station included three to four different activities, where children were given a prop (e.g., a puzzle), and asked to perform an action-based behavior (e.g., Please solve the puzzle). The order of the stations and activities within each station was randomized between participants.

To reduce memory demands, children were interviewed about the three to four activities they completed after each station (rather than at the end of all 17 activities). Following each activity station, the experimenter excused herself from the room, and the interviewer entered the room to ask the child questions about their own actions in the previously completed activities. The experimenter left all the objects used in the previous activities on the table to act as visual cues and allow the interviewer to point to the exact activity they were referring to during questioning. Participants were randomly assigned to one of two question-type conditions: Why (i.e., “Why did you put the missing piece in the puzzle?”), or How Come (i.e., “How come you put the missing piece in the puzzle”). The question-type condition remained constant across all activity stations. One question was asked per activity and questions were asked in the same order as the participant experienced the activities. The interviewer then exited the room and the experimenter returned. This procedure was repeated until all five activity stations were complete.

Finally, all children completed the Woodcock–Johnson IV (WJ IV) test of oral language picture vocabulary task (Schrank et al., 2014) to assess children’s oral language development and lexical knowledge. Children were presented with a series of pictures and asked to label each picture. The task was terminated when six consecutive pictures were incorrectly labelled. Children received one point for each correctly identified picture. Scores can range from 0 to 54.

Coding How–Why Task Responses

Children’s responses to the interviewer’s questions were coded as one of three, mutually exclusive, codes: accurate (explained a cause for their behavior related to the task, demonstrating an understanding of the intention of the question; e.g., I did it because she asked me to/because I wanted to), inaccurate (e.g., failed to comprehend the question, and in turn explain the cause of their behavior; e.g., “good,” “bad,” “easy,” “fun,” etc.), or IDK. We then calculated a sum for each type of response outlined below and divided these scores by the total number of participants’ responses (17), creating proportion scores of accurate, inaccurate, and IDK responses for each child. Inter-rater reliability was assessed using a random sample of participants, evenly distributed across condition and age, for 10% of the sample (κ > 0.90).

Results

Accurate Responses

First, we were interested in whether children’s accuracy increased with age. We also assessed the influence of question type to determine whether children were significantly less accurate when answering Why questions compared to How Come questions. To assess children’s accuracy, a 2(Condition: Why, How Come) × 3(Age: 5, 7, 9) univariate analysis of variance (ANOVA) was performed on the proportion of children’s accurate responses out of 17 (see Table 1 for means and SDs). There was a significant main effect of Age, F(2,114) = 14.22, p < .001, η_p² = .20. Planned contrasts (Bonferroni) revealed that, as predicted, 5-year-olds (M = .65, SD = .25) were significantly less likely to provide accurate responses, compared to 7-year-olds (M = .85, SD = .25, p = .001, 95% confidence interval [CI: −.33, −.07]) and 9-year-olds (M = .93, SD = .25, p < .001, 95% CI [−.42, −15]). There was no significant difference between 7-year-olds and 9-year-olds, p = .422, 95% CI [−.21, .05]. There was no main effect of Condition, F(1,114) = 1.52, p = .220, η_p² = .01 (Why M = .84, SD = .24; How Come M = .79, SD = .24) or interaction between Age and Condition, F(2,114) = 2.96, p = .056, η_p² = .05.

Table 1.

Mean (Standard Deviation) Proportion Scores for Accurate, Inaccurate, and “I don’t Know” Responses, by Age and Condition.

Condition	Accurate	Inaccurate	I don’t know
5-year-olds
Why	.71 (.25)	.12 (.21)	.17 (.17)
How Come	.59 (.25)	.24 (.21)	.17 (.17)
7-year-olds
Why	.93 (.25)	.02 (.21)	.05 (.17)
How Come	.78 (.25)	.19 (.21)	.03 (.17)
9-year-olds
Why	.89 (.25)	.06 (.21)	.05 (.17)
How Come	.98 (.25)	.02 (.21)	.00 (.17)

Note. Age is between-subjects, including children 5 (n = 40), 7 (n = 40), and 9 years (n = 40).

Inaccurate Versus IDK Responses

Given that 7- and 9-year-olds were highly accurate (85% and 93%, 7- and 9-year-olds, respectively) when answering Why/How Come questions, we only examined 5-year-olds’ responses when they failed to provide an accurate response. A 2(Condition: Why, How Come) × 2(Response Type: Inaccurate, IDK) repeated-measures ANOVA was performed with the proportion of responses as the dependent variable. There was no main effect of Condition, F(1,38) = 1.29, p = .263, η_p² = .03, or Response Type, F(1,38) = .02, p = .882, η_p² = .001, indicating that 5-year-olds were equally likely to provide inaccurate (M = .18, SD = .26) and IDK responses (M = .17, SD = .28). The interaction was not significant, F(1,38) = .62, p = .436, η_p² = .02.

Mediation Analysis—The Mechanism of Language Influencing Accuracy

Prior to testing the mediation model, correlations were performed to examine the relationship between age (5-, 7-, 9-year-olds), language (WJ IV scores), and accuracy (proportion of accurate responses). Results indicated that all variables were significantly positively related, meeting the requirements for a mediation analysis (see Table 2 for correlations).

Table 2.

Correlations between Age, Language, and Accuracy.

Variable	1	2	3
1. Age	—
2. Language	.71*	—
3. Accuracy	.52*	.54*	—

Note. Correlations between Age and Language, and Age and Accuracy calculated using Spearman’s rho, n = 120.

^*p < .001.

A mediation analysis was performed with age as the predictor, language as the mediator, and accurate responses as the outcome variable. All path coefficients among age, language, and accuracy were significant (see Figure 1). The total effect of age on accurate responses was significant, R² = .190, F(2,119) = 13.70, p < .001. However, the direct effect of age on accuracy, controlling for language, was not significant, R² = .006, F(2,116) = .49, p = .613, showing evidence that language mediates the relationship of age and accurate responses. See Table 3 for path weights and 95% CIs.

Figure 1.

Mediation Model of the Relationship Between Age and Accuracy With Language as the Mediator (n = 120). Each pathway includes the percent of variance explained in the direct relationship of one variable to another (R²). The c path represents the total effect of age on accuracy. The c′ path represents the direct effect of age on accuracy after accounting for language. *p < .001.

Table 3.

Unstandardized Effects for the Relationship Between Age, Language, and Accuracy.

		95% CI
	Point Estimate	Lower	Upper
Total effect of age on accuracy
5 years	.65	.57	.73
7 years	.20	.09	.31
9 years	.28	.17	.39
Direct effect of age on accuracy
5 years	−.01	−.34	.31
7 years	.06	−.06	.18
9 years	.06	−.08	.21
Age on language
5 years	22.38	21.40	23.34
7 years	4.85	6.03	8.77
9 years	7.40	3.48	6.22
Language on accuracy	.03	.02	.04

Note. Age is between-subjects, including children 5 (n = 40), 7 (n = 40), and 9 years (n = 40). The 95% CIs that do not include zero indicate significance.

Discussion

The current study was the first to investigate the accuracy of children’s responses to Why and How Come questions. Overall, children’s accuracy improved with age, highlighting a developmental milestone during the elementary school years in which children become near ceiling in accuracy for Why and How Come questions, and there were no differences between question type. Furthermore, the relationship between age and accuracy was mediated by children’s language, suggesting that linguistic development plays a role in children’s ability to answer Why and How Come questions.

The Developing Understanding of Why/How Come Questions

To gain further insight into the development of children’s abilities to accurately answer Why and How Come questions, we examined children’s age in relation to their performance. In line with our predictions, findings revealed that in response to Why and How Come questions older children (7- and 9-year-olds) were significantly more likely to provide accurate responses, compared to younger children (5-year-olds). These findings suggest that there is a drastic developmental shift between 5 years of age and 7 years of age when children become better able to answer Why and How Come questions, and this accuracy becomes near ceiling by 9 years of age.

A possible explanation for children’s improvement with age is their increased familiarity with and exposure to Why and How Come questions. Malloy and colleagues (2016) suggest that younger children often experience greater difficulties with Why questions, as they require an understanding of concepts (e.g., causation) for which young children are less familiar. Furthermore, past research has widely established that children’s lack of experience may undermine their ability to accurately provide responses to later acquired wh- questions, including Why questions (Cameron-Faulkner et al., 2003; Rowland et al., 2003; Valian & Casey, 2003). Taken together, our findings, along with previous studies, indicate that improvement with age is possibly driven by older children’s greater experience with Why and How Come questions.

Importantly, while children gain experience with age, they also develop a greater metalinguistic awareness of intentionality in communication exchanges (Edwards & Kirkpatrick, 1999). Therefore, it is also possible that children’s improved performance with age is driven by a greater linguistic understanding of the intentions for Why and How Come questions. In fact, our findings revealed that language mediated the relationship between age and accurate responses, suggesting that with age children’s oral language and lexical knowledge improves, which in turn increased children’s ability to provide accurate responses to Why and How Come questions. Notably, this finding highlights the important role that language plays in children’s ability to answer adults’ questions.

Children’s Accuracy

While previous studies have examined the productivity of children’s responses when answering Why and How Come questions (e.g., Ahern et al., 2018; Andrews et al., 2016), this study is the first to examine the accuracy of children’s responses. Results revealed that children were quite accurate when answering Why and How Come questions, though this dramatically developed with age. These findings suggest that children not only provide productive responses (as seen in Andrews et al., 2016) but also accurate responses to commonly used questions.

Furthermore, findings revealed no significant differences in children’s accuracy when answering Why or How Come questions. Although past studies have combined these two questions in analyses, as there are no differences in the productivity of children’s responses (see Andrews et al., 2016), research also suggests that semantic differences in question forms can influence how the question is interpreted, and in turn the accuracy of children’s responses (see Evans et al., 2014; Wylie et al., 2019). Therefore, given the differing interpretations of Why (e.g., accusatory and critical) and How Come questions (e.g., surprise and disbelief), we examined these questions separately, with the goal of informing best practices for phrasing questions to young children. However, findings from this study suggest that the semantic differences in Why and How Come questions did not influence children’s ability to accurately describe the cause of their behavior.

Notably, the current study was comprised of positive events (e.g., completing a puzzle), which may have minimized the accusatory nature of questioning. Furthermore, this study provides a foundation for understanding children’s accuracy when responding to Why and How Come questions. However, future research is needed to examine how children respond within more applied paradigms, such as a more forensically relevant environment (e.g., longer time delays; negatively valenced events). Future studies should also examine children’s ability to reason about other’s actions, given that the current study focused solely on children’s ability to describe the cause of their own behaviors.

Children’s Inaccurate Responses

In addition to examining the accuracy of children’s responses, we were also interested in examining children’s response patterns when failing to answer Why and How Come questions. That is, while older children most often provided accurate responses to Why and How Come questions, 5-year-olds demonstrated clear difficulties, failing to provide an accurate response 35% of the time. Furthermore, when 5-year-olds failed to provide an accurate response, they were just as likely to provide an inaccurate (18% of the time) or IDK response (17% of the time). Although IDK response can be effective in expressing a lack of ability to answer the question, inaccurate responses are more problematic. In this study, children’s inaccurate responses represent a lack of understanding for the intention of the question, as their responses entirely failed to address a cause of their behavior. In line with past research (Saywitz et al., 1993; Waterman et al., 2000), children are attempting to answer a question that they do not understand, potentially leading to miscommunications between the adult and child.

Conclusion

Taken together, the present study provides insight into the development of children’s abilities to accurately answer Why and How Come questions. Furthermore, these findings build on the growing body of literature examining developmentally appropriate questioning practices, in determining questions that result in miscommunications or misunderstandings between children and those questioning children. Children’s accuracy significantly improved between 5 years of age and 7 years of age and became near ceiling at 9 years of age, following the pattern of children’s productivity noted across studies by Malloy and colleagues (2016) and Andrews and colleagues (2016), and this developmental shift was largely explained by children’s improved language abilities. However, there were no differences in children’s accuracy when responding to Why or How Come questions. Ultimately, these findings help to inform individuals working with children (e.g., parents, teachers, social workers, researchers, legal professionals) on how to inquire about children’s positive experiences, suggesting that the age of the child as well as children’s language repertoire, rather than the semantics of Why and How Come questions, should be taken into consideration when inquiring about their event knowledge.